The overall goal of the following experiment is to demonstrate the method for multi objective optimization of conservation practices in a watershed using a simulation optimization system involving the watershed process model, and an evolutionary algorithm. The overarching question is how to allocate agricultural conservation practices in a watershed. So water quality objectives are met at the lowest cost.
Multiple conservation practices are possible on each field, and multiple water quality objectives may be important. A particular assignment of conservation practices can be simulated by a watershed process model. To achieve the optimization goal first, select a calibrated and validated watershed process model and model representations of conservation practices.
As a second step, the environmental objectives to be maximized are selected and costs of conservation practices are obtained, which allow for the evolutionary algorithm component looking to simultaneously optimize along environmental and cost objectives to be invoked. Next, selection of parameters controlling optimization is done in order to perform multi objective optimization. These two components, simulation and optimization are integrated into a simulation optimization system called genetic I swat results are obtained that show the optimal set of watershed configurations in terms of conservation practice placements, which quantifies the trade-offs between environmental goals and the cost of conservation investments, and allows selection of a specific spatial configuration of conservation practices based on desired environmental objectives or cost.
The main advantage of this technique over the existing methods like simple conservation practice scenario assessment or optimization of the choice of conservation practices based on simplified practice representation, is that it integrates physically based watershed process model into the optimization decision in a way that is flexible and intuitively understandable. This method can help answer key questions in the watershed management and environmental economic fields, such as where to focus public investments in conservation practices, or how to structure market-based policies such as reverse auctions for conservation practices or a water quality trading program. In the context of non-point source pollution Optimization parameters are selected after a watershed model is prepared and input data for optimization provided optimization is controlled by a program called Genetic Iwo.
To begin this procedure, open genetic iwo dot exe go to file, then open and select the I SWAT database, raccoon GA dot mdb go to file, then configuration to assign the paths to swap. Model executables go to execute, then select allele set. This step determines the combinations of conservation practices used in optimization.
For this run allele set number 14 will be used, which has 23 combinations of conservation practices go to execute. Then select SP A two archive baseline aware subset to perform multi objective optimization using the SP A two evolutionary algorithm. First, under preset, select the watershed to be optimized raccoon clicking apply selects entries from the presets file watershed presets dot csv to fill control values on this screen.
Next, under output variable, select the environmental objectives for optimization. The selected N outlet P outlet defines a three-dimensional objective function. Nitrogen averaged for five years at the outlet, phosphorus averaged for five years at the outlet and the total cost of conservation practices.
This will create a three-dimensional trade-off frontier set the initial population size at 60. This determines the initial number of candidate solutions when the seed with each allele option is selected. Candidate solutions representing a uniform application of each conservation practice specified in the allele set to all cropland hydrologic response units in the watershed are created.
First, the remaining candidate solutions are created by a random assignment of conservation practices from the allele set to cropland HR.When selecting the seed with each allele option, make sure that the initial population size, which is 60 in this demonstration, is at least as large as the number of alleles in an allele set, which is 23. In this demonstration, set the desired number of generations or iterations for the optimization run in this example to 125. When two candidate solutions are selected for creating new candidate solutions, crossover probability specifies the probability that distinct new solutions are created.
For this demonstration, the crossover probability is set to one. The size of temporary population determines the number of new candidate solutions created. Processor resources are used most efficiently when this value is an integer, multiple of the number of processor threads 16 is selected for this demonstration.
The mutation probability is the probability of random change in HRU assignment to another conservation practice. From the allele set, it is set to 0.003. For this demonstration, select the number of threads or processors used, which is 16.
In this demonstration, the curve number calibration factor of one is provided from the swap model calibration. Finally, select save population in text file. Checking this option produces a text file with the allele values of every HRU in every surviving candidate solution.
This is important for restarting the optimization run after the specified number of iterations is completed. After the run, the entire set of Pareto efficient solutions or the trade-off frontier can be visualized by following these steps. Run genetic.
I swat, go to file, then open to open the I IWA database, raccoon GA dot mdb. Go to file export, then export HRU list, save file as raccoon allele, HRU dot T XT run map, swat dot xe, select execute, and then 3D animation to produce animation of the three dimensional trade-off frontier, that pits outlet nitrogen levels N on the red axis against outlet phosphorus levels P on the blue axis against the summed cost of conservation practices. In all subbasins.
On the green axis, output is a series of files which can be rendered all at once into image files. By using the POV ray program, the images can also be combined into a movie showing the algorithm progression by running frame scanner xe. Each point in the frontier represents a watershed configuration.
That is a specific assignment of conservation practices on a landscape. Many of these configurations can be seen for the entire frontier by following these steps. Run, map, swat, xe, select execute, and then map animation.
The boxes on the left side show the two dimensional projections of the frontier and the dash lines denote the position of the particular solution chosen. The MAP shows the dominant algorithm prescribed conservation practice. In each where the legend identifies the chosen conservation practices.
Often a question of interest is to select a specific watershed configuration or an individual achieving a specified set of water quality objectives. For example, an individual reducing nitrogen by 30%and phosphorus by 30%relative to baseline loadings map. SWAT allows us to search the frontier for an individual with the minimum EUCLIDEAN distance to the specified objective To select specific watershed configurations or individuals achieving particular water quality objectives, open map swat dot exe, and select execute and search.
Enter a minimum target zero. In this example, a maximum target 100 in this example, as well as a target interval 10. In this example, enter a specific percent reduction in nitrogen from baseline in the percent reduction space next to end baseline 30.
In this example. Then enter percent reduction in phosphorus in the percent reduction next to phosphorus baseline, also 30. In this example, the map swat program will produce output in a pop-up screen, click copy, text, and paste into a spreadsheet.
Three tables are produced in the first are individuals nearest to N and P targets of the same percent reduction, which ranges from T in to tmax by t int. Just below this, the closest single individual to the target N spec and P spec appears in the second table are nearest individuals, where the P target ranges from team in to Tmax while N is held constant near end spec. The third table gives individuals nearest end targets ranging from team in to Tmax while P is held constant near P spec.
In this case, the closest individual to a 30%end reduction was ID 84 23 with an end value of 14, 639, 660. Here is the map showing the spatial distribution of conservation practices and the location of this watershed configuration in the trade-off frontier After its development. The technique paved the way for researchers in the field of watershed management and environmental economics to explore more cost-effective ways of achieving watershed environmental objectives and to improve the design of market-based policies.
